New Three-Way Split Mold Design and Experimental Procedure ... · New Three-Way Split Mold Design...

Chadi S. El Mohtar1 and Dennis A. Rugg2

New Three-Way Split Mold Design and ExperimentalProcedure for Testing Soft, Grouted Soils

ABSTRACT: Soil grouting has become a popular method for soil improvement in recent years. Grouting is generally intended to increase asoil’s strength, increase its liquefaction resistance, or reduce its hydraulic conductivity. Soil grouting involves the injection, permeation, or me-chanical mixing of cementitious, silica, or clay grout into a soil deposit. With the increase in popularity of these methods comes the issue of testinggrouted soils to verify the expected soil improvement. While some of the methods and materials mentioned result in soils that are sufficientlycemented to produce trimmable specimens that can stand under their own weight, other methods produce softer materials that are very difficult tosample or even to recreate and test in the lab. Preparing such soft samples in the laboratory poses two challenges: 1) if the specimen is prepared inthe triaxial cell directly, the grouting process might not be feasible because of the porous stones and small diameter tubing in the triaxial cell; and2) if the specimens are prepared outside the triaxial cell, soft specimens might not be able to stand under their own weight without significantstrains and damage to the soil structure. This paper will thoroughly describe a three-way split mold specifically designed to accommodate the per-meation and testing of soils that are too soft or too weak to be easily sampled or tested in the lab. A simple procedure outlining the use of thisthree-way split mold will also be described. Finally, the results from a series of consolidated undrained, monotonically loaded triaxial tests will bepresented as an example of the split mold application. These tests utilized the new three-way split mold for sample preparation of loose Ottawasand permeated with a thixotropic bentonite suspension.

KEYWORDS: grouting, split mold, consolidated undrained triaxial, liquefaction, soil improvement

Nomenclature

Cc ¼ coefficient of curvatureCO2 ¼ carbon dioxide

CS ¼ critical stateCU ¼ consolidated undrained monotonic triaxial tests

with pore pressure measurementsCu ¼ coefficient of uniformity

D10 ¼ diameter with 10% of the soil finerD50 ¼ diameter with 50% of the soil finer (average

diameter)D60 ¼ diameter with 60% of the soil fineremax ¼ maximum void ratioemin ¼ minimum void ratioGs ¼ specific gravity of solidsLL ¼ liquid limit

NCEER ¼ National Center for Earthquake EngineeringResearch

PL ¼ plastic limitPVC ¼ polyvinyl chloride

R ¼ stress ratioRCS ¼ stress ratio at CS

RUIS ¼ stress ratio at UIS

SPP ¼ sodium pyrophosphateUIS ¼ undrained instability state

USCS ¼ Unified Soil Classification System(ea %) ¼ axial strain

/0 ¼ effective internal friction angle/0CS ¼ effective internal friction angle at CS/0UIS ¼ effective internal friction angle at UIS

q ¼ deviatoric vertical stress (r1�r3)p0 ¼ mean effective stress [1/2](r01þr03)

Introduction

Liquefaction of loose, saturated sands poses a major threat to infra-structure worldwide due to the large deformations observed whenthis phenomenon occurs. Many soil improvement methods havebeen introduced to counteract this threat, the easiest of which is soilcompaction by different methods. However, soil compaction (espe-cially dynamic compaction) can often be harmful to existing struc-tures on or nearby liquefiable soil deposits. Thus, permeationgrouting or “passive site remediation” has been proposed for stabi-lizing liquefiable soil deposits at sensitive or developed sites. Pas-sive site remediation techniques, as described by Gallagher (2000),involve the penetration of various flowable grouts into the porespaces of a liquefiable soil deposit, under low pressures, such thatthe initial soil fabric is not significantly disturbed.

Researchers have proposed the use of many different groutingmaterials, including cement based grouts (e.g., Dano et al. 2004;Dupla et al. 2004), colloidal silica grouts (e.g., Maher et al. 1994;Liao et al. 2004; Gallagher et al. 2007; Gallagher and Lin 2009),

Manuscript received November 30, 2010; accepted for publication May 20,2011; published online July 2011.

1Assistant Professor, Dept. of Civil, Architectural & EnvironmentalEngineering, Univ. of Texas at Austin, Austin, TX 78712. (Correspondingauthor), e-mail: [email protected]

2Geotechnical Engineer, Golder Associates, Lakewood, CO 80228, e-mail:[email protected]

Copyright VC 2011 by ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. 1

Geotechnical Testing Journal, Vol. 34, No. 6Paper ID GTJ103645

Available online at www.astm.org

Copyright by ASTM Int'l (all rights reserved); Wed Nov 16 18:00:15 EST 2011Downloaded/printed byUniv of Texas Austin pursuant to License Agreement. No further reproductions authorized.

and clay grouts (Haldavnekar et al. 2003; El Mohtar et al. 2008).Cement and colloidal silica grouts tend to produce soils that aresufficiently strong to allow for trimming and testing under uncon-fined conditions. Clay grouts; on the other hand, can producesofter soil specimens that may not be able to stand under theirown weight. This creates an issue when trying to verify theimproved soil’s parameters. These soils are nearly impossible tosample and very difficult to create and test in the lab.

Verifying the properties of the improved soil is essential todescribe the effectiveness of the utilized grout. In 1997, the NCEERmodified and updated a simplified procedure by Seed and Idriss(1982) for assessing liquefaction resistance by correlating the stand-ard penetration blow counts or the cone penetration resistance to aliquefaction potential. However, Gallagher (2000) states that thiscorrelation may not be adequate for grouted soils since there is notmuch data to make the correlation. Thus laboratory testing must beimplemented. The preparation of laboratory samples permeatedwith cement or colloidal silica grout is not a trivial matter; however,these soil mixtures are generally adequately cemented to allow foreasy trimming and sample preparation after permeation and curing.Specimens permeated with clay suspensions can be more difficultto trim and prepare in the lab since they are softer and generallyrequire constant confinement.

A method for permeating and testing soft, grouted, cohesion-less soils will be presented in this paper. Tests were run specifi-cally on loose, saturated sand permeated with thixotropicbentonite suspensions. The equipment and procedures describedherein could be used for testing a wide variety of soft, permeatedsoils under many different conditions.

Background

Many of the available soil improvement methods have been dis-cussed in depth by Welsh (1987), Van Impe (1989), Hausmann(1990), Broms (1991), Bell (1993), Mosely (1993), and Kramer(1996). These authors discuss various topics, including densifica-tion, reinforcement, drainage, and grouting. Much of the discus-sion on grouting methods involves the injection, permeation, ormechanical mixing of cementitious, silica, or clay grouts. Theimproved performance imparted by cement or silica is due to irre-versible chemical bonds (Maher et al. 1994; Gallagher and Mitch-ell 2002). These bonds cannot be recovered once broken, butthese materials result in soils that are adequately cemented andthus easily sampled, trimmed, and tested.

Dano et al. (2004) investigated the use of cement grouts for soilstabilization. The authors prepared specimens of two differentsands: Fontainebleau sand and Seine River alluvial sand. The speci-mens were prepared in 80 or 100 mm diameter PVC rigid tubes thatwere 900 mm long. The sand was air pluviated into the tubes withzero drop height and then either flushed with water or kept dry.Afterwards, the sand was permeated with cement grout from bottomto top at a constant flow rate of 180 mL/min. After permeation, thecolumns were cured in a temperature and humidity controlled roomfor 28 days. Following the curing period, the cemented sand wasextruded from the tube and four specimens were trimmed fromeach column to be tested either in unconfined compression ordrained triaxial shear. Zebovitz et al. (1989), Benhamou (1994),

and Schwarz and Krizek (1994) have all used similar specimenpreparation methods to prepare cement grouted sands.

Dupla et al. (2004) investigated permeation and stabilizationusing cement-bentonite grout. The permeation columns were 80 mmin diameter and 1040 mm in height (five 200 mm sections). The col-umns were made out of plexiglass and included 5 pore pressuretransducers equally spaced along the entire length. A 40 mm thicklayer of gravel was placed at the bottom of the column to preventsand from entering the injection port and to allow for a uniform dif-fusion of the grout into the sand. The sand was air pluviated into thecolumn, flushed with CO2, and then flushed with water. The groutingwas performed with a diaphragm pump with an adjustable flow rate(0–600 mL/min). The grouted columns were allowed to cure underwater for 28 days. After the curing period, the column was trimmedinto 5 specimens and unconfined compression tests were performedon each specimen. The common 28-day resting period for the Duplaet al. (2004) and Dano et al. (2004) studies is a result of ASTM C39/C39M, which specifies a 28-day strength for concrete testing (andmany other international standards). Both investigations producedsoil specimens that were cemented, easily trimmed, and were suffi-ciently strong to be tested unconfined.

Gallagher and Lin (2009) permeated columns of Nevada orOttawa sand with colloidal silica grout. The permeation setup con-sisted of three 300 mm sections made of 100 mm internal diametertransparent PVC aligned on top of each other. Each of the three sec-tions was split vertically to facilitate removing the sand post grouting.The bottom section was assembled on the bottom platen and a metalfilter screen was used at the bottom of the column to prevent the inletchannel from clogging. A 45 mm thick gravel layer was placed ontop of the metallic screen and the sand was then placed through waterpluviation to fill the first section. The second two sections were thenassembled one at a time following the same procedure. The specimenwas then flushed with tap water for 2–3 days before grouting. Fol-lowing grouting, the columns were allowed an initial gel time, afterwhich the stabilized columns were trimmed, covered in plastic wrapand aluminum foil, and stored for “at least 4 times the initial geltime.” Again, these specimens were strong enough to be trimmedand tested unconfined. The authors concluded that the measuredunconfined compressive strengths of the specimens (40–60 kPa)were high enough “to be able to mitigate the liquefaction risk.”

Haldavnekar et al. (2003) and El Mohtar et al. (2008) haveinvestigated the use of thixotropic bentonite clay suspensions toreduce the liquefaction susceptibility of clean sands. The sand andbentonite were dry mixed and air pluviated into a triaxial speci-men, then flushed with CO2 and water. The specimens wereallowed enough time for the bentonite and water to form a gel-likestructure that contributes to the cyclic strength of cohesionlesssoils. This approach was first used to overcome the complicationsof grouting sands with bentonite grouts. Conversely to the easilytrimmed, cemented structure created by permeation with cementi-tious or silica grouts, the grouted sand created by permeation withbentonite clay grouts is soft and not cemented. This results in soilthat cannot easily be sampled or trimmed for triaxial testing, andis too soft to stand on its own without continuous confinement, letalone being tested for unconfined compressive testing.

Finally, El Mohtar (2008) investigated the use of bentoniteclay suspensions as a permeation grout material. This study

2 GEOTECHNICAL TESTING JOURNAL


utilized an early three-way split mold design that allowed for thepermeation of a 72 mm diameter sand specimen. The three-waymold could then be partially disassembled for trimming into a tri-axial specimen while keeping the central portion of the specimensupported by the inner split mold.

The current paper describes a modified and updated three-waysplit mold designed specifically for permeating specimens thatrequire confinement during trimming and triaxial specimen prepa-ration. The following sections will describe, in detail, the moldand its dimensions and the procedure used for permeation andtrimming. The designed mold allows for the placement of filtermaterials at the top and bottom of the sand specimen to allow fora uniform grouting front through the whole specimen length. Themold also applies some confinement to the specimen throughoutthe preparation process up to placement in the triaxial setup.Finally, the results of consolidated undrained monotonic triaxialtests (CU) on loose, saturated Ottawa sand permeated in the newthree-way split mold will be presented. These tests will indicatethe repeatability and uniformity of specimens produced with thisnew three-way split mold and its effectiveness in maintaining con-tinuous confinement to the grouted sand.

Development of the Three-Way Split Mold

All current literature on preparing grouted sand specimens focuseson using some form of a rigid wall column permeation setup andrelies on the grouted sand having enough strength to be extrudedout of the column, trimmed, and tested. However, none of thedocumented methods can be used to prepare non-cementedgrouted specimens since all the methods require removing thespecimens out of the permeation mold before testing. Therefore,there is a need to develop a new permeation mold that providessome confinement throughout grouting, storing, trimming, andsample preparation. In addition, the new mold should allow for acoarse material filter at the bottom (and top if needed) to provide auniform grout front through the final specimen. The following

sections describe the development of a new three-way split moldthat was specifically designed for preparing triaxial specimens thatwere permeated with non-cementing grouts. This mold allows forthe creation of a soil specimen that will not leave the central splitmold (similar to the split mold traditionally used for preparingsand triaxial specimens) until it is securely placed on the base ofthe triaxial cell under a vacuum.

First, the conceptual design, sketch, and first generation moldwill be presented, followed by a detailed description of the secondgeneration mold and the final design drawings. The details of thesample preparation method will be described for loose, saturatedOttawa sand specimens that were permeated with a thixotropicbentonite suspension. This procedure could be further modifiedfor testing soils permeated with various grouting materials.

Schematic Design and First Generation Mold

Figure 1 shows a schematic of the conceptual design that both thefirst and second generation molds were based upon. The three-way split mold consists of 5 main pieces: the top plate, the topsplit mold, the central split mold, the bottom split mold, and thebase plate. The base and top plates need to be fitted with an inflowand an outflow valve, respectively, for flushing the specimenswith CO2, water, and finally grout. The top, central, and bottomsplit molds should be split along the direction of grouting. The topand bottom split molds will be dissembled after grouting is com-plete in order to trim out the excess sand/filter material. Mean-while, the central split mold will not be dissembled until thespecimen is placed in the triaxial cell and is under vacuum. Thissequence ensures that the grouted sand remains under a minimumamount of confinement from the end of sand placement in themold up to the end of the triaxial test. All of the pieces should beprecisely machined and fitted with the proper O-rings to create aholistic device that is air and water tight to approximately 400kPa.

The central split mold hosts the final triaxial specimen inside thelatex membrane wrapped around the top and bottom of the mold.The top and bottom split molds allow for placing a coarse materialfilter and an additional layer of the same sand to be tested. Thisincreases the uniformity of the grouted sand within the central splitmold by (1) having a more horizontal grout front penetrating thesand because of the higher hydraulic conductivity of the coarser fil-ter material; and (2) through trimming out any caking that mightoccur at the interface between the filter material and the tested sand.

As described previously, the first generation three-way splitmold design was proposed by El Mohtar (2008) (see Fig. 2(a)).The original design utilized a typical triaxial split mold, two off-the-shelf pipe repair clamps (Fig. 2(b)), two plexiglass plates, andfour threaded rods. The two plexiglass plates were fitted with inletand outlet valves and used as the top and bottom plates. The piperepair clamps were placed on the top and bottom one-third of thesplit mold; the rubber interior allows for forming a seal betweenthe clamps and both the central split mold and the top and bottomplexiglass plates. While the pipe repair clamps did not split intotwo separate pieces, loosening the clamps allowed for a largeenough gap between the central split mold and the pipe repairclamps to dissemble the setup for trimming. The threaded rods

FIG. 1—Schematic of the three-way split mold.

EL MOHTAR AND RUGG ON NEW THREE-WAY SPLIT MOLD FOR GROUTED SOILS 3


were used to hold the five pieces together. While this designproved the possibility of using the three-way split mold to prepare“undisturbed” soft grouted triaxial specimens, the design requiredmany upgrades to facilitate the process and improve the quality ofthe final specimen.

Second Generation Mold

Figure 3 shows a detailed design drawing of the second generationthree-way split mold (referred to as split mold for the remainder ofthe paper). Figure 3 shows half of the assembled mold (middle),

FIG. 2—First generation three-way split mold from El Mohtar (2008).

FIG. 3—Detailed design drawing of the second generation three-way split mold.



plan view of the top cap (top right), plan view of the top splitmold (top left), plan view of the central split mold (bottom left),and plan view of the bottom split mold (bottom right). The piperepair clamps were replaced with true top and bottom split moldsand the base plate was incorporated into the bottom split mold toprovide a more rigid and stable system. All of the parts (except forthe threaded rods and bolts which were made out of steel) weremade of 6061 Aluminum since it is soft enough for precisemachining while strong enough to sustain the expected pressures.

The top cap is a 120.7 mm square and has a thickness of 12.7mm. The center of the top cap is fitted with a threaded hole for thesplit mold’s exit valve. Additionally, four 6.5 mm diameterunthreaded holes were drilled at the four corners for the threadedrods to pass through for the final assembly.

The top split mold was machined out of a rectangular alumi-num block that has a 120.7 mm square base and is 78.7 mm thick.A 78 mm diameter cylinder was drilled at the center of the squareface of the block throughout its whole thickness. The hollowblock was then split vertically into two identical parts. Each halfof this split mold was fitted with two 4.8 mm thick aluminumdowel pins on one side and two 4.8 mm diameter pinholes on theother. These pins were placed 19.5 mm (one/fourth the length ofthe top split mold) from the top and bottom of the split mold toallow for alignment of the two halves when assembled. Four 76.2mm long hex-screws (6.4 mm diameter) were used to hold the twohalves together after assembling. On one of the halves, four 6.4mm threaded holes were drilled to tighten the screws into, whileon the other half a 6.5 mm unthreaded hole was drilled throughoutthe width of the section for the screw to pass through. This holewas doubled in size for the first 38 mm to allow for the wholescrew to be embedded inside the mold. The upper face of the topsplit mold has an O-ring groove with an outer diameter of 88.9mm machined into it. The groove is 2.7 mm wide and 1.5 mmdeep. The O-ring in this groove forms a seal between the top capand the top split mold. Finally, four 6.5 mm diameter unthreadedholes were drilled vertically at the four corners of the split moldfor the threaded rods to pass through for the final assembly.

The central split mold is similar to any standard split moldused for preparing sand triaxial specimens with two major differ-ences to accommodate the specifics of its use:

(1) It has a much thinner wall than a typical triaxial split mold.The thinner walls lead to smaller inside diameter of the topand bottom split molds, which in turn minimizes theamount of material that needs to be trimmed later on. Moreimportant, the thinner walls cause minimal change in theseepage velocity of the grout when its front reaches thecentral split mold by reducing the change in cross-sectionalarea under the constant flow rate.

(2) The height to diameter ratio is equal to 2. The grouted sandinside the central split mold will end up forming the triax-ial specimen, and therefore, needs to have a height to diam-eter ratio of 2. Standard triaxial split molds usually have alower portion that wraps around the bottom platen of thetriaxial cell to provide stability and an airtight seal betweenthe membrane and the mold. In the setup presented here,the membrane is folded around the top and bottom of the

split mold to form the tight seal for applying vacuum later(see Fig. 6, to be presented later).

The central split mold is 3.2 mm thick with an inside diameterof 71 mm and a height of 142 mm. Each of the two halves has a1.6 mm aluminum dowel pin and a 1.6 mm hole placed at 35 mmfrom the top and bottom ends. These pins and holes ensure theproper alignment of the central split mold during assembly. Twohose clamps are used to hold the central split mold together duringthe final setup. In the center of each half of the central split mold isa fitting for a vacuum line. These fittings are used to apply vacuumbetween the membrane and the mold in order to allow for the mem-brane to take the same shape as the central split mold. A 1.9 mmwide and 1.2 mm deep O-ring grove was machined 25.4 mm fromeach end of the central split mold. The O-rings used in this groovehelp to create a leak-free seal between the central split mold and thetop and bottom split molds. The 25.4 mm space between the end ofthe mold and the O-ring will have the latex membrane rolled overit until the specimen is placed in the triaxial cell.

The bottom split mold was machined out of a rectangular blockof aluminum with a 120.7 mm square base and a 95.3 mm thick-ness. Again, a 78 mm diameter cylinder was hollowed out at thecenter of the square side of the block to a depth of 79.6 mm. The78 mm diameter was calculated to account for the outer diameterof the central split mold (77.4 mm) and twice the thickness of thelatex membrane that will be rolled over the central split mold (0.6mm). The hollow block was then split vertically into two identicalparts. Each half of this split mold was fitted with two 4.8 mm thickaluminum dowel pins and two 4.8 mm diameter pinholes. Eachhalf of the bottom split mold was also machined with four holesfor the locking screws in a similar configuration as described forthe top split mold. Four 6.4 mm threaded holes (two in each halfof the split mold) were drilled vertically at the four corners; thesefour holes are aligned with the unthreaded four holes at the cor-ners of the top split mold and the top cap. Two threaded rods (6.4mm diameter) are rigidly attached to each half of the bottom splitmold; these rods run through the top split mold and the top capand are used to support the top split mold as well as to tighten thetop cap in place. Finally, the solid base of the bottom split moldwas drilled with a 10.2 mm diameter drainage hole to be fittedwith an inlet/flushing valve. First, a vertical hole was drilled at thecenter of the hollow cylinder created in the bottom split mold;then this hole was continued in the horizontal direction to the sideof one of the halves.

Figure 4 shows a photograph of the partially assembled split mold,displaying all of its parts except for the top cap. The top and bottomsplit molds are not completely assembled to show the alignment pins,holes, and threaded holes to hold the final assembly together. Alsoseen in Fig. 4 is the vacuum inlet to help stick the latex membrane tothe inside of the central split mold and the inflow valve used to flushthe sand with CO2, water, and the grout. With the threaded rodsscrewed into the bottom split mold, the assembled three-way splitmold can stand on its own even when partially assembled.

While the first generation mold utilized readily available mate-rials to create a mold that adequately performed the desired func-tion, there was a need to create a holistic device designedspecifically for permeation and triaxial specimen preparation. For



the first generation mold, the maximum grouting pressure waslimited because of various leaks that occurred when the pressuresexceeded 75 kPa. This was mainly due to difficulties in consis-tently assembling the setup since the different components werenot originally designed to be used together. The second generationmold was very precisely machined, sturdier than the first genera-tion mold, and easier to work with due to its square, stable design.The square cross section of the top and bottom split molds, alongwith the alignment provided by the four threaded rods, allow foreasier trimming of the top and bottom of the specimen and, there-fore, reducing the disturbance to the final grouted specimen insidethe central split mold. Additionally, the second generation moldallows for using higher grouting pressures in excess of 450 kPa.

Testing Procedures

Split Mold Assembly

The aluminum, three-way split mold is a unique device and as-sembly must be done carefully to help assure that all joints aretight and leak free. The following gives a detailed outline of thesteps taken to assemble and utilize the mold effectively.

Step I—Both halves of the upper and lower portions of thesplit mold are assembled separately on the threaded rods as seen inFig. 5. The upper and lower split mold halves with holes for thelocking screws must be on the same set of threaded rods. One nuton each threaded rod can be adjusted to mark the location of theupper split mold. The space between the bottom and top split moldscan be adjusted to increase or decrease the amount of filter materialand sand needed in the top and/or bottom split molds. The onlyrestriction on this spacing is to keep it short enough so that the O-rings at the top and bottom of the central split mold will still overlapwith the top and bottom split molds to ensure a tight seal. Once thedesired length is established, the contact areas of the split moldhalves are covered with a thin layer of vacuum grease to accountfor scratches and irregularities and to assure a good seal.

Step II—The central split mold is cleaned, the joints arelightly greased, the two halves are aligned, and the mold isadequately tightened by two steel hose clamps. Three O-rings arethen stretched on both the top and bottom of this mold. The firstO-ring on each side is placed in the O-ring groove to provide aseal between the central split mold and each of the top and bottomsplit molds. A thin layer of grease is applied over these O-rings tofurther ensure the integrity of the seal. The other two O-rings oneach side will be used later to seal the specimen to the top and bot-tom platens of a triaxial cell. For these two O-rings, the centralsplit mold is used as an O-ring stretcher to avoid any disturbanceto the specimen when placing the O-rings later on.

Step III—The latex membrane (0.3 mm thickness) is cut pre-cisely to a length such that, when stretched over the interior of thecentral split mold and rolled over the outer edge of the mold, itreaches just short of the O-rings, as seen in Fig. 6. Because of the0.6 mm difference between the inner diameter of the top and bottomsplit molds and the outer diameter of the central split mold, therolled over portion of the membrane will not be damaged during thefinal assembly. Vacuum is applied to the central split mold to helpthe latex membrane assume the shape of the mold. Care is taken sothat the membrane is not twisted or over-stretched within the centralsplit mold since these conditions can cause unnecessary andunknown stresses on the specimen once the vacuum is removed.

Step IV—The central split mold is positioned on the half ofthe assembly discussed in the first step that contains the threadedholes for the locking screws as seen in Fig. 6. The central splitmold is placed such that the O-rings fall approximately 5 mmwithin the edges of the top and bottom split molds and the vacuumlines are on a horizontal plane.

Step V—The second half (with the unthreaded holes for thelocking screws) is placed on top of the first half of the split mold

FIG. 5—Step I of assembly—connect the two halves of the top and bottom splitmolds.

FIG. 4—Partially assembled three-way split mold (except for top cap).



(with the inlet valve facing down) and the locking screws areinserted and tightened as shown in Fig. 7. At this point, the fullyassembled mold (except for the top cap) is stood up and ready tocontain a new specimen. The top cap will be placed after pluviat-ing the soil into the mold.

Specimen Preparation

As described previously, this three-way split mold is designed totest cohesionless soils permeated with non-cementing grouts. Theresulting material is usually insufficiently strong to stand under itsown weight and requires a minimum amount of confinementthroughout specimen preparation. The procedure described in thefollowing paragraphs outlines the steps to prepare grouted sandspecimens using the three-way split mold for triaxial tests. This

procedure can and should be slightly altered to accommodate thetesting of different materials under different conditions.

First, the filter material is placed in the bottom third of the splitmold (about 40 mm thick). The filter material is added to aid inthe permeation of the specimen. It allows the bentonite suspensionto propagate horizontally and vertically forming a uniform frontbefore reaching the targeted sand. The top of the filter materialshould be at least 10 mm below the bottom edge of the centralsplit mold. The presence of this gap will allow the grouted sand tostart below the central split mold; thus allowing for trimming outany caking that might occur at the bottom of the grouted sand andcreate a more uniform specimen within the central split mold.

After the placement of the filter materials, sand is air-pluviatedwith a funnel. The drop height of the sand was controlled by theoperator and, as a result, there was a slight variation in the relativedensities of each specimen. This user controlled drop heightbetween the funnel and the top of the sand/filter material in themold can be adjusted, along with the funnel diameter, dependingon the desired final relative density; higher free drop height ofsand and a larger diameter funnel results in denser specimens. Thesand is pluviated through the central mold and into the top splitmold; the sand should extend at least 10–20 mm beyond the topof the central split mold for trimming purposes. Finally, a layer offilter material (approximately 10 mm thick) is placed at the top ofthe specimen to prevent sand from piping out through the effluenttube (see Fig. 8). Next, an O-ring is placed in the groove on theupper face of the top split mold and is covered in a thin layer ofvacuum grease. The top cap is then secured in place on thethreaded rods with four nuts. A photograph of the completelyassembled split mold is shown in Fig. 9.

The specimen is then slowly flushed with CO2. Following theCO2 flushing stage, the specimen is saturated with de-aired waterat a slow rate not to disturb the sand structure. Approximately twopore volumes of de-aired water are flushed through the specimen.At this stage, the specimen is close to saturation and ready to be

FIG. 7—Step V of assembly—tightening both halves of the top and bottom splitmolds.

FIG. 8—Three-way split mold with specimen and filter material in (just beforeplacing top cap).

FIG. 6—Steps II–IV of assembly—completely assembled central split moldadjusted in place.



grouted. A schematic of the grouting setup can be seen in Fig. 10.The specimen is grouted using a peristaltic pump with varyingflow rates up to 100 mL/min. The pressure is monitored at thebase of the mold during permeation. Prior to permeation, careshould be taken to ensure that all lines connected to the specimenare saturated. The specimen is then grouted.

Following successful grouting, the influent and effluent lines tothe split mold are shut and the specimen is allowed a period of rest.Grouting was considered to be successful when one full pore vol-ume of grout (or more) was permeated into a specimen utilizingless than 250 kPa of pressure. This rest allows for the grout to buildup a minimal strength within the pore spaces, creating a material

that can be easily trimmed. This time should vary depending on thegrout used. After the rest period, the split mold is disassembled.First, the top cap is removed and then the split mold is rested on itsside with the locking screws facing upward (similar to Fig. 7). Thelocking screws are then removed and the upper halves of the topand bottom split molds are taken off. A completely permeated, par-tially disassembled specimen can be seen in Fig. 11.

Next, the filter materials are removed from the top and bottomsplit molds along with part of the grouted sand. Approximately0.5–1 cm of the grouted sand is left protruding from the centralsplit mold before it is removed from the remaining halves of thetop and bottom split molds (Fig. 12(a)). If the length of the soilextending beyond the mold is too long, this could cause negativetensile stresses at the top and cause cracking or breakage of thespecimen. A steel wire saw is then used to trim the remaining per-meated sand from the central split mold. Special care should betaken to protect the membrane during the trimming stage so that it

FIG. 10—Schematic of permeation setup using the peristaltic pump.

FIG. 9—Completely assembled split mold.

FIG. 11—Partially dissembled grouted specimen with half of the bottom filtermaterial trimmed.



is not sliced or punctured. After the specimen is adequatelytrimmed, as seen in Fig. 12(b), the membrane is thoroughlycleaned to remove any sand particles or grout that might compro-mise the integrity of the membrane’s seal against the top and bot-tom platens of the triaxial cell. The specimen is now ready to beplaced in the triaxial cell.

A saturated filter paper and a saturated porous stone are thenplaced on the triaxial base platen. The specimen is stood up verti-cally and carefully placed on the base platen, while still inside thecentral split mold (with vacuum still applied between the moldand membrane) as seen in Fig. 13. The top filter paper and porousstone are then placed on top of the specimen and the top platen isaligned and set in place. A thin layer of vacuum grease is placedon the top and bottom platens before placing the specimen in thetriaxial cell to help ensure a tight seal between the membrane andthe platens. Next, the membrane is rolled up onto to the top platenand down onto the base platen and the O-rings are then rolled intoplace. A vacuum is applied to the specimen through the top andbottom pore pressure lines prior to removing the central splitmold. This vacuum puts the specimen under minimal confiningpressure as the central split mold is disassembled. The triaxial cellis then completely assembled and the testing performed in a simi-

lar manner to any standard triaxial test. After the completion ofthe test, the final void ratio was calculated from measuring thetotal mass, total volume, water content, and bentonite content ofthe specimen. The initial void ratio at the end of grouting wasbackcalculated from the recorded volume changes during the dif-ferent testing stages.

Preliminary Results

The main purpose of this section is to show that the three-waysplit mold can be used to test uncemented grouted sands thatwould not be possible to test with the previous specimen prepara-tion techniques found in the literature. These tests were performedto determine the effect of bentonite grouts on the static liquefac-tion of a loose fine sand. Only preliminary results are presentedhere since the main scope of the paper is to introduce the newspecimen preparation method.

Materials

Ottawa sand (ASTM graded sand C778) was utilized for thesetests as the sand to be grouted. A summary of Ottawa sand proper-ties can be seen in Table 1. Approximately 20 mm of fine

FIG. 12—Central split mold (a) with 10 mm of permeated sand left, and (b) af-ter trimming.

FIG. 13—Specimen mounted on the triaxial cell while still confined by the cen-tral split mold.

TABLE 1—Summary of Ottawa sand properties.

Gs 2.65 Cu 1.61

emin 0.50 Cc 1.07

emax 0.76 D10 (mm) 0.23

USCS SP D60 (mm) 0.37



aggregate (Pea stone, diameter > 4.75 mm) and 20 mm of coarsesand (1.2 mm< diameter< 1.75 mm) were used as the two layersfor the bottom filter materials. These filter materials were selectedbased on the grain size distribution of the Ottawa sand and to sat-isfy Terzaghi’s retention criteria for filters.

Wyoming sodium-bentonite grade CP-200 was used to preparethe grouts for these tests. A summary of the bentonite clay proper-ties can be found in Table 2. Grouts with 7.5% and 10% bentoniteby total mass of the grout were prepared using high shear mixing.Since these grouts were too viscous to be permeated through thefine sand selected, the grouts were modified using SPP. Sodiumpyrophosphate has a specific gravity of 1.8. The chemical structureof the SPP is Na4P2O7�10H2O and the molecular weight is 446.06.The pH of a 5% solution was measured to be 9.5 using a JENCO60 pH meter. The SPP was selected because, while it significantlyreduces the initial viscosity of the bentonite suspensions, the viscos-ity is recovered over time (Clarke 2008). 1% and 2% SPP by drymass of bentonite were added to the 7.5% and 10% bentonitegrouts, respectively. De-ionized de-aired water was used for thepreparation of all bentonite grouts and for all triaxial tests.

Specimen Preparation Specifics

The specimen preparation method was similar to the one listed inthe previous sections. To avoid repetition, the specifics of thespecimen preparation are listed in the order of occurrence duringthe process:

(1) During Ottawa sand pluviation, the funnel was keptapproximately 5 mm from the top of the sand to produceloose sand specimens (relative density of 35%–40%).

(2) Following the air pluviation of the sand specimens, CO2

was slowly flushed through the specimen; CO2 is used sinceit dissolves more readily in water at low pressures and roomtemperature than regular air and will help achieve higherdegrees of saturation later on during the backpressure stage.After flushing with CO2, two pore volumes of de-aired andde-ionized water were flushed through the specimen at aflow rate of approximately 6 mL/min. This allowed thespecimens to be as close to saturation as possible (withoutthe use of backpressure) prior to permeation.

(3) The specimens were grouted using a peristaltic pump at arate of 20 mL/min. The pressures were monitored at thebase of the mold during grouting and the test was termi-

nated if the pressures reached 250 kPa. The completegrouting took around 25 min.

(4) After grouting, the specimens were allowed 48 h restingtime before starting the disassembling process.

(5) After placement in the triaxial cell, the specimens weresubjected to -25 kPa pore pressure before removing thevacuum between the membrane and the central split moldand dissembling the central split mold. The negative porepressures were applied from the top and monitored fromthe base using a pore pressure transducer to insure that thevacuum was applied uniformly across the specimen.

(6) The triaxial cell was then completely assembled and a cellpressure of 25 kPa was applied while removing the negativepore pressure and maintaining a constant effective stress.Specimens were then backpressure saturated to achieve a B-value of at least 0.95 before being consolidated to the finaleffective confining stress of 150 kPa. It generally tookbetween 200 and 300 kPa of backpressure to reach thedesired saturation values. Finally, the specimens were shearedundrained with pore pressure measurements (CU tests).

For the clean sand specimen, the sand was air pluviated intothe triaxial split mold with a 5 mm free drop to generate a loosespecimen with a relative density between 35% and 40%. Thespecimen was kept under 25 kPa confining pressure while beingflushed with CO2 and de-aired water. The sand was then backpres-sure saturated to reach a B-value of 0.95 or higher before beingconsolidated to 150 kPa and sheared undrained. More details onthe clean sand specimen and grouted specimen preparation can befound in Rugg (2010). By using similar pluviation methods forpreparing the sand specimens and the grouted specimens, theeffects of the grout can be better compared.

Permeation Results

Table 3 shows some of the pertinent information regarding threepermeated specimens. These specimens were permeated in thethree-way split mold with either a 10% bentonite suspension (with2% SPP) or a 7.5% bentonite suspension (with 1% SPP). As seenin Table 3, the 10% bentonite suspensions required maximum per-meation pressures near 175 kPa, while the 7.5% bentonite suspen-sions only required approximately 30 kPa maximum permeationpressure. The skeletal relative densities of these specimens werebetween 35% and 40% and the total permeation time was on theorder of 25–30 min. The large change in permeation pressuresbetween the 10% bentonite suspensions and the 7.5% bentonite sus-pensions can be explained by the difference in rheological proper-ties described in the following paragraph and displayed in Fig. 14.

Figure 14 shows the change in yield stress over time for 7.5%and 10% bentonite suspensions with and without SPP. The yield

TABLE 2—Properties of bentonite (modified from Mitchell and Soga 2005).

PL 35 % CEC 80–150 meq/100 g

LL 190 %–1160 % Specific area 700–800 m2/g

Initial water content 8.3 % Swelling capacity 8 mL/g

TABLE 3—Representative permeation results for various grouted specimens (modified from Rugg 2010).

Specimen Suspension Max pressure, kPa Permeation time, min:s Bentonite content, % DR, SK, %

20091202 10B2SPP 173 30:24 2.27 36.9

20091209 10B2SPP 172 23:17 2.39 35.4

20100709 7.5B1SPP 31 23:40 1.71 40



stress of the 7.5% and 10% bentonite suspensions decreases signifi-cantly with the addition of 1% and 2% SPP, respectively. The yieldstress of the suspensions with SPP show a much higher thixotropicincrease in the yield stress over the first 24 h compared to that ofthe untreated bentonite suspensions; eventually, the yield stress ofthe suspensions with SPP exceeds that of the untreated suspensions.This significant reduction in the initial yield stress increases the mo-bility of the suspensions and allows for successful permeation.

Table 4 shows the bentonite content at the top, middle, and bot-tom of each of the specimens described in Table 3. The averagebentonite content in the specimens permeated with the 10% benton-ite suspensions is around 2.3% by dry mass, while the average ben-tonite content in the specimen permeated with a 7.5% bentonitesuspension is approximately 1.7%. Table 4 shows that a slight fil-tration effect occurs along the length of the specimen. It can beseen that the bentonite content at the base of each specimen isslightly higher than the bentonite content at the top of each speci-men. Figure 15 shows the grout pore pressure versus time duringthe permeation process. The linear increase in pressure with timeindicates that little to no caking occurs during the permeation pro-cess. The bentonite content was assumed to be uniform across thehorizontal cross section of the specimen and the boundary had noeffect on the bentonite flow through the specimen. The sand used isa fine sand with d50¼ 0.35 mm and dmax¼ 0.7 mm. The diameterof the specimen is 71 mm (100 times the maximum particle diame-ter); therefore, the boundary effects can be neglected.

Undrained Triaxial Test Results

The results from four undrained monotonic triaxial tests are pre-sented in this section. The four tests were run on a clean sandspecimen, a specimen permeated with 7.5% bentonite and 1%

SPP (7.5B1SPP) and two specimens permeated with 10% benton-ite and 2% SPP (10B2SPP and 10B2SPP-2). Figure 16 shows thedeviatoric vertical stress (r1–r3) versus the axial strain (ea %) forthe three tested specimens. The clean sand specimen shows a typi-cal behavior with the deviatoric stresses increasing up to a localpeak (undrained instability state or UIS as defined by Murthyet al. 2007) and then starts dropping, showing a tendency to liq-uefy and flow. Since the tested specimens all had a skeletal rela-tive density (for the sand specimen, the skeletal and bulk relativedensity are the same) of 35%–40%, the sand specimen did notundergo static liquefaction, but rather started dilating when theaxial strains exceeded 2%. This phase transformation from con-tractive to dilative behavior can be seen in the excess pore pres-sure generation plot shown in Fig. 17. The positive excess porepressures start decreasing when the axial strains exceed 2% andnegative pore pressures are generated at higher strains.

The permeated specimens did not show any UIS (Fig. 16), butrather a slow change in the slope of the stress-strain curve. This

TABLE 4—Bentonite content at the top, middle, and bottom of three perme-ated specimens.

Bentonite content, %

Specimen 20090709 20091202 20091209

Top 1.68 2.09 2.37

Middle 1.68 2.32 2.37

Bottom 1.77 2.42 2.41

FIG.15—Grouting pressure versus time.

FIG. 16—Deviatoric stress versus axial strain from the CU tests.

FIG. 14—Rheological properties for 10% and 7.5% bentonite suspensions.



can be easily explained by the lower positive excess pore pressuresgenerated during shearing of the permeated specimens and themore rapid phase transformation to a dilative behavior. This morerapid phase change results in the generation of negative pore pres-sures at much lower strains (3% and 5% axial strain as comparedto 12% axial strain for the clean sand). This more rapid phasetransformation can be clearly seen in Fig. 18 as well; the figureshows the q–p0 plot for the three specimens. The clean sand plotshows the tendency for static liquefaction with the curve propagat-ing left before the phase transformation and having p0 and qincreasing simultaneously. The permeated specimens showed amuch reduced tendency for static liquefaction with higher benton-ite concentration in the grouts; the plot for the grouted sand is tothe right of the clean sand plot. This change in response understatic loading, from the limited results presented in this paper, indi-cates a higher resistance to static liquefaction due to the reductionin the generation of positive pore pressures.

Table 5 shows the stress ratio (R¼ r01/r03) values and equiva-lent secant friction angle at the UIS and CS for the three speci-mens. For the grouted sands, since there is no true UIS, thelocation of the initial change in slope on the stress-strain plots wasused to calculate RUIS. The critical state values are calculated at

higher strains and can be backcalculated from the final slopes inFig. 18. The clean sand and grouted specimens show similar shearstrength properties despite the introduction of the grout into thepores. These results, in particular the UIS results, which occur atsmall strains (less than 0.5% axial strain), indicate that there is lit-tle change to the sand structure during grouting as well as duringspecimen preparation and installation into the triaxial cell. Thesecombined results clearly show the applicability of using the three-way split mold to prepare uncemented, cohesionless, grouted sandtriaxial specimens without damaging the structure of the sand.

Summary and Conclusions

A new three-way split mold was developed for preparing triaxialspecimens of uncemented grouted sands. The new mold consists ofthree inter-connected split molds that allow for preparing a uni-formly grouted specimen by having a filter material at the base andtop that can eventually be trimmed out when preparing the finalspecimen to be tested. This new specimen preparation procedureprovides a minimal confinement throughout the specimen prepara-tion stages; this is critical to preserve the soil structure when thegrouted soil is not strong enough to stand alone. Finally, prelimi-nary tests on sand specimens grouted with bentonite suspensionswere presented to show a possible application for the new mold.The results showed a reduction in the tendency of the sand to liq-uefy under static loading in the presence of the bentonite grout inthe pore space. These results also showed that the grouting did nothave a significant effect on the shear strength of the sand, indicatingthat the specimen preparation method was successful in retainingthe original sand structure. This hypothesis is further justified bythe similar secant friction angles at low strains for both clean andgrouted sands. While the uniformity of permeation grouting in thefield would remain a big concern, the proposed mold and testingprocedures open the door for laboratory testing on soft groutedspecimens that was not feasible before.

Acknowledgments

The writers would like to acknowledge the Civil, Environmentaland Architectural Engineering Department at the University ofTexas at Austin for the financial support provided to Mr. Ruggduring the period of this project.

References

ASTM C39/C39M, “Standard Test Method for CompressiveStrength of Cylindrical Concrete Specimens,” Annual Book ofASTM Standards, ASTM International, West Conshohocken, PA.

Bell, F.G., 1993, Engineering Treatment of Soils, E&FN Spon,London.FIG. 18—Stress path of the tested soils.

TABLE 5—Stress ratio and internal friction angle.

RUIS u0UIS RCS u0CS

Clean sand 2.11 21� 2.88 29�

7.5B1SPP 2.20 22� 3.12 31�

10B2SPP 2.19 22� 3.11 31�

FIG. 17—Excess pore pressure versus axial strain from the CU tests.



Benhamou, O., 1994, “Comportement rheologique des coulis deliants hydrauliques ultrafins destines al’injection [Rheologicalbehavior of ultrafine cement slurry intended for injection],”Ph.D. thesis, Ecole de Geologie de l’Ingenieur, Ecole Natio-nale Superieure des Mines de Paris (in French).

Broms, B., 1991, “Deep Compaction of Granular Soil,” Founda-tion Engineering Handbook, 2nd ed., H.-Y. Fang, Ed., VanNorstrand Reinhold, New York, pp. 814–832.

Clarke, J., 2008, “Investigation of Time-Dependent RheologicalBehavior of Sodium Pyrophosphate - Bentonite Suspension,”Masters thesis, Purdue University, West Lafayette, IN.

Dano, C., Hicher, P., and Tailliez, S., 2004, “Engineering Proper-ties of Grouted Sands,” J. Geotech. Geoenviron. Eng., Vol.130, No. 3, pp. 328–338.

Dupla, J., Canou, J., and Gouvenot, D., 2004, “An AdvancedExperimental Set-Up for Studying a Monodirectional GroutInjection Process,” Ground Improvement, Vol. 8, pp. 91–100.

El Mohtar, C. S., 2008, “Pore Fluid Engineering: An Autoadap-tive Design for Liquefaction Migration,” Ph.D. thesis, PurdueUniversity, West Lafayette, IN.

El Mohtar, C. S., Clarke, J., Bobet, A., Santagata, M., Drnevich,V., and Johnston C., 2008, “Cyclic Response of a Sand withThixotropic Pore Fluid,” Geotechnical Special Publication, n181, 2008, Proceedings of the Geotechnical Earthquake Engi-neering and Soil Dynamics IV Congress 2008-GeotechnicalEarthquake Engineering and Soil Dynamics, GeotechnicalSpecial Publications (GSP) 181, Vol. IV.

Gallagher, P., 2000, “Passive Site Remediation for Mitigation ofLiquefaction Risk,” Ph.D. dissertation, Virginia PolitechnicInstitute and State University, Blacksburg, VA.

Gallagher, P., and Lin, Y., 2009, “Colloidal Silica Transportthrough Liquefiable Porous Media,” J. Geotech. Geoenviron.Eng., Vol. 135, No. 11, pp. 1702–1712.

Gallagher, P., and Mitchell, J., 2002, “Influence of ColloidalSilica Grout on Liquefaction Potential and Cyclic UndrainedBehavior of Loose Sand,” Soil Dyn. Earthquake Eng., Vol. 22,No. 9-12, pp. 1017–1026.

Gallagher, P., Pamuk, A., and Abdoun, T., 2007, “Stabilization ofLiquefiable Soils Using Colloidal Silica Grout,” J. Mater. Civ.Eng., Vol. 19, pp. 33–40.

Haldavnekar, V., Bobet, A., Santagata, M., and Drnevich, V.P,2003, “Soil Treatment with Thixotropic Fluid: An Autoadap-tive Design for Liquefaction Prevention,” Proceedings of the11th International Conference on Soil Dynamics and Earth-quake Engineering and 3rd International Conference onEarthquake Geotechnical Engineering, Vol. II, Stallion Press,pp. 553–560.

Hausmann, M. R., 1990, Engineering Principles of Ground Modi-fication, McGraw-Hill, New York.

Kramer, S. L., 1996, Geotechnical Earthquake Engineering, Pren-tice Hall, Upper Saddle River, NJ.

Liao, H., Huang, C., and Chao, B., 2004, “Liquefaction Resist-ance of a Colloidal Silica Grouted Sand,” Grouting andGround Treatment, ASCE, Reston, VA.

Maher, M., Ro, K., and Welsh, J., 1994, “Cyclic UndrainedBehavior and Liquefaction Potential of Sand Treated withChemical Grouts and Microfine Cement (MC-500),” Geotech.Test. J.,Vol. 17, No. 2, pp. 159–170.

Mitchell, J., and Soga, K., 2005, Fundamentals of Soil Behavior,3rd ed., Wiley, New York.

Mosely, M. P., 1993, Ground Improvement, Blackie, London.Murthy, T., Loukidis, D., Carraro, J., Prezzi, M., and Salgado, R.,

2007, “Undrained Monotonic Response of Clean and SiltySands,” Geotechnique, Vol. 57, No. 3, pp. 273–288.

Rugg, D., 2010, “Undrained, Monotonic Shear Strength of LooseSaturated Sand Treated with a Thixotropic Bentonite Suspen-sion for Soil Improvement,” Master’s thesis, The University ofTexas at Austin, Austin, TX.

Schwarz, L. G., and Krizek, R. J., 1994, “Effect of PreparationTechnique on Permeability and Strength of Cement-GroutedSand,” Geotech. Test. J., Vol. 17, No. 4, pp. 434–443.

Seed, H. B., and Idriss, I. M., 1982, Ground Motions and SoilLiquefaction During Earthquakes, Earthquake EngineeringResearch Institute, Berkeley, CA.

Van Impe, W. F., 1989, Soil Improvement Techniques and TheirEvolution, Balkema, Rotterdam, The Netherlands.

Welsh, J.P., Ed., 1987, “Soil Improvement: A Ten Year Update,”Geotech. Spec. Publ. No. 12.

Zebovitz, S., Krizek, R. J., and Atmatzidis, D. X., 1989, “Injectionof Fine Sands with Very Fine Cement Grout,” J. Geotech.Geoenviron. Eng., Vol. 115, No. 12, pp. 1717–1733.



http://dx.doi.org/10.1680/grim.2004.8.3.91

http://dx.doi.org/10.1061/(ASCE)GT.1943-5606.0000123

http://dx.doi.org/10.1061/(ASCE)GT.1943-5606.0000123

http://dx.doi.org/10.1016/S0267-7261(02)00126-4

http://dx.doi.org/10.1061/(ASCE)0899-1561(2007)19:1(33)

http://dx.doi.org/10.1061/(ASCE)0899-1561(2007)19:1(33)

http://dx.doi.org/10.1520/GTJ10088J


http://dx.doi.org/10.1680/geot.2007.57.3.273


New Three-Way Split Mold Design and Experimental Procedure ... · New Three-Way Split Mold Design...

Documents

Transcript of New Three-Way Split Mold Design and Experimental Procedure ... · New Three-Way Split Mold Design...